The objective is to develop new regression models for analyzing clustered observations, as found in familial data, where the individuals in a cluster are correlated and the outcomes are categorical. These models will be studied using data from the Pima Indian and Bedouin Arab infant feeding studies.